Whenever I review a schematic or a PCB layout for a younger engineer, =
I make=20
a point to pick out items and simply ask them =E2=80=9Cwhy?=E2=80=9D Why =
did you choose this=20
component? Why did you place it here on the PCB? My point in asking =
these=20
questions is that every design decision should have a valid reason that =
the=20
engineer understands.
For example, a resistor in the feedback path of an op amp configured =
as a=20
buffer provokes an immediate =E2=80=9Cwhy?=E2=80=9D
Figure 1: An op amp buffer circuit with a resistor =
included in=20
the feedback path
The ugly truth is that often an engineer doesn=E2=80=99t know why he =
or she included=20
resistor R2. They may have seen it on a previous schematic and felt =
obliged to=20
include it. Most commonly, these resistors are used in low speed =
applications=20
(<50 MHz) to cancel the DC offset produced by the input bias current =
of the=20
op amp. However, as my now-retired colleague Bruce Trump pointed out, this is rarely =
effective.
R2 may also provide some protection to the inverting input in the =
event of an=20
ESD strike at the output. Also, some op amps, especially JFET input =
types, may=20
produce lower distortion if both inputs have matched source impedances. =
But=20
without understanding the purpose of R2, the value is often selected =
randomly,=20
leading to instability.
To understand how R2 causes instability, let=E2=80=99s add the input =
capacitance of=20
the op amp to the circuit as shown in Figure 2.
Figure 2: Op amp input capacitances added to the =
circuit of=20
figure 1
R2 forms a low pass filter at the inverting input with the input =
common-mode=20
capacitance CCM2 and the input differential capacitance=20
CDM. The low-pass filter in the feedback path produces a zero =
in the=20
noise gain curve (1/=CE=B2) of the amplifier at the frequency:
If this zero occurs far above the unity gain bandwidth of the =
amplifier, as=20
shown by fz(2) in Figure 3, it should not affect the =
stability=20
of the circuit. However, if the zero occurs at or below the unity gain =
bandwidth=20
of the op amp, like fz(1), the noise gain curve will =
intersect=20
the open loop gain curve at greater than 20dB/decade, indicating =
probably=20
instability.
Figure 3: Open loop gain (red) and noise gain (blue) =
curves=20
for a buffer amplifier
The phase shift from this zero will begin 1 decade below=20
fz, so a conservative design guideline is:
By substituting the equation for fz, =
we can=20
determine the maximum value for R2 to ensure stability:
To show this effect, I simulated an OPA172 in a buffer configuration and measured the =
phase=20
margin for different values of R2. The unity gain bandwidth of the OPA172 =
is 10MHz and=20
the input common-mode and differential capacitances are both 4pF. Using =
the=20
design guideline in equation 4, the maximum value for R2 is:
The TINA-TI=E2=84=A2 simulation schematic I used to =
measure phase margin=20
is shown in figure 4. The feedback loop is broken at the op amp output =
by=20
inductor L1 and a source (VG1) is AC coupled into the feedback loop. The =
loop=20
gain is given by the probe labeled =E2=80=9CLG=E2=80=9D and phase margin =
is measured when the=20
loop gain =3D 0dB.
Figure 4: TINA-TI=E2=84=A2 simulation schematic for an OPA172 configured as a buffer
Figure 5 is a plot of the phase margin as the value of R2 =
is=20
increased. The blue line is the maximum value of R2 we =
calculated in=20
equation 5. Below this limit, the reduction in phase margin is =
minimal,=20
with it falling to 62=C2=B0 at R2 =3D 200 Ohms. Above this =
limit, the phase=20
margin decreases rapidly.
Figure 5: Plot of phase margin versus R2 =
value for=20
the OPA172
Remember that this analysis doesn=E2=80=99t include the effects of =
capacitive loads=20
or PCB parasitics, which will also degrade the phase margin of the =
circuit.
In some circuits there may be a valid reason to include R2, but ask =
yourself=20
what you=E2=80=99re trying to accomplish with this resistor before =
putting it in the=20
schematic. If the required value is large, you=E2=80=99re probably =
headed for stability=20
problems. Asking yourself =E2=80=9Cwhy=E2=80=9D in any engineering =
pursuit is crucial to=20
improving as an engineer!